1. Field of the Invention
The present invention relates to a digital video signal processing device, and, more particularly, to a digital video signal processing device which employs a digital filter.
2. Description of the Related Art
In general, a video signal contains a large amount of information per unit time. The amount of information in a video signal per unit time increases greatly when the video signal is digitized. For example, when a NTSC television signal is converted to an 8-bit digital signal at a sampling frequency of 4 fsc (fsc: a color subcarrier frequency), a resultant digital signal has an amount of information which corresponds to a transmission bit rate of about 120 Mbps. Further, when three R, G, and B signals of the above-described NTSC television signal are respectively converted to 8-bit digital signals at a sampling frequency of 3 fsc, resulting digital signals exhibit a transmission bit rate of about 260 Mbps.
In recent years, research has been conducted on high-definition television signals obtained using a number of horizontal scanning lines per field which is about twice that for conventional television signals. A digital high-definition television signal has a large amount of information per unit time, and the transmission bit rate thereof is extremely high. For example, when R, G and B video signals of a high-definition television signal are respectively converted to 8-bit digital signals at a sampling frequency of 64 MHz, the bit rate for resultant signals is 1.5 Gbps.
In order to improve the transmission efficiency, so-called time compressed integration (TCI) has been practiced recently in which a chrominance signal is time-base compressed relative to a luminance signal, the resultant chrominance signal being time-division multiplexed with the luminance signal. Even when a high-definition television signal on which time compressed integration has been performed is converted to an 8-bit digital signal with a sampling frequency of 64 MHz, a resultant digital signal is transmitted at a bit rate as high as about 510 Mbps.
Thus, a digital video signal has a high bit rate, and transmission of such a digital signal through a telecommunication system therefore requires a telecommunication line of large capacity, thus making the capacity of telecommunication lines inadequate and/or increasing transmission costs. Further, recording of such a digital signal having a high bit rate on a recording medium such as a magnetic recording medium creates certain problems since recording heads are incapable of coping with such a signal or recording times are shortened.
Accordingly, various bandwidth compression techniques have been proposed for the purpose of reducing the amount of information contained in a digital signal. One of them is the so-called offset subsampling technique. In this offset subsampling technique, the components of an image in diagonal directions which are not important from the visual point of view are removed by a spatial filter, and high-frequency components are provided in this cleared frequency region so as to lower the sampling frequency. Such an offset subsampling falls into two categories: one is a line offset subsampling (LOSS) in which sampled positions shift in adjacent scanning lines in one field, and the other is a field offset subsampling (FOSS) in which sampled positions shift in adjacent scanning lines of adjacent fields. In both cases, two-dimensional band limitation has to be conducted using the above-described spatial filter which acts as a prefilter before offset sampling is performed. Further, when the digital signal on which offset subsampling has been performed is decoded, spatial interpolation has to be conducted using a spatial filter which acts as a postfilter.
In the case of subsampling a color video signal, subsampling is performed on each of the component signals, e.g., a luminance signal and two types of color-difference signals, which means that spatial filters serving as a prefilter and postfilter is required for each signal.
These spatial filters generally include a plurality of series-connected one horizontal scanning period (H) delay lines, and a plurality of multipliers for multiplying respective coefficients, and are relatively large in size. Conventional offset subsampling which is one form of bandwidth compression technique requires provision of a plurality of number of spatial filters, and this increases the overall size of a device and production costs.
If a spatial filter is of the type which has 25 taps for multiplying 25 picture elements including a center picture element by coefficients, as shown in FIG. 1, the number of coefficient multipliers, multipliers, adders and digital delay lines in a signal processing circuit must be made to correspond to the number of taps. Each of the coefficient multipliers may be a RAM or ROM table. As a result, if offset subsampling is performed on a magnetic recording/reproducing machine such as a VTR, a spatial filter having a large configuration must be respectively provided for recording and reproduction, increasing the overall size of a filter circuit. Similarly, in the case of a transmitter/receiver for telecommunication lines, a spatial filter is required for both of the transmitting and receiving sides, thereby increasing the overall size of a circuit. In FIG. 1, reference symbols x and y denote gaps between the picture elements in the horizontal and vertical directions, respectively.